kkomyoeminaung/qwen2.5-14b-linear-merged

TEXT GENERATIONConcurrency Cost:1Model Size:14.8BQuant:FP8Ctx Length:32kTool Calling:SupportedPublished:Jun 1, 2026Architecture:Transformer Gated Cold

The kkomyoeminaung/qwen2.5-14b-linear-merged model is a 14.8 billion parameter language model created by kkomyoeminaung using a linear merge method. It combines the base Qwen2.5-14B-Instruct with Qwen2.5-Coder-14B-Instruct, emphasizing enhanced coding capabilities. This model is designed for general instruction-following tasks with a strong focus on code generation and understanding, leveraging its 32768 token context length.

Loading preview...

Model Overview

The kkomyoeminaung/qwen2.5-14b-linear-merged is a 14.8 billion parameter language model developed by kkomyoeminaung. It was created using the Linear merge method via MergeKit.

Key Capabilities

This model is a blend of two foundational models from the Qwen family:

  • Qwen/Qwen2.5-14B-Instruct: Serving as the primary base model with a weight of 0.8, providing strong general instruction-following abilities.
  • Qwen/Qwen2.5-Coder-14B-Instruct: Integrated with a weight of 0.2, specifically enhancing its performance in code-related tasks.

This strategic merge aims to combine robust general language understanding with specialized coding proficiency, making it suitable for a variety of applications requiring both.

Use Cases

Given its composition, this model is particularly well-suited for:

  • General instruction following: Responding to diverse prompts and carrying out various language tasks.
  • Code generation: Producing code snippets, completing functions, and assisting with programming challenges.
  • Code understanding and analysis: Interpreting existing code, explaining logic, and potentially debugging assistance.
  • Applications requiring a balance of general AI and coding expertise.